Document 6717724

advertisement
1.
Introduction
Eternalism is undeniably a very popular view in metaphysics1. But there is really no single view that
the name ‘eternalism’ reliably picks out; rather, there is a cluster of views and the conjunction of
some or all of these is variously designated by ‘eternalism’. In what follows we pull apart this cluster
of views to present a more nuanced characterisation of the various different versions of eternalism that
one finds.
More specifically, the structure of this paper is as follows. We begin by characterising the
ontological foundations of eternalism, providing a precise characterisation of the eternalist’s
ontological commitments. We then consider the role that the so-called temporal B-series plays in
eternalism, using the B-series to differentiate between a number of variations on the core eternalist
position. We then go on to consider the fundamentality of the B-series. That is, we discuss whether or
not the eternalist should think that the B-series is fundamental. We finish by briefly considering the
relationship between presentism and eternalism. We present a new argument for the view that the
ontological picture advocated by the eternalist is to be preferred to the ontological picture endorsed by
presentists.
2.
The Ontic Component
Call the ontological foundations of eternalism the ontic component. As a first pass, the ontic
component of eternalism can be stated as follows:
Ontic Component (1): The past, present and future exist (unrestrictedly).
It is our view that the ontic component, appropriately understood, is an undeniably essential
component of eternalism. The use of the modifier unrestrictedly in the above is to draw attention to
the fact that ‘exists’ is being used to pick out the unrestricted existential quantifier rather than a
quantifier that is restricted to, say, a particular time or place. For everyone agrees that all and only the
events and objects that exist now, exist now. The central idea behind this characterisation of the ontic
component then is that while neither the past nor the future exist now, they do exist, simpliciter. In
this sense the past and future are just like other locations in space: just as Hong Kong does not exist
here, (since we are in Sydney) and yet it does exist. This characterisation is supposed to set eternalism
apart from a view like presentism, according to which although the past did exist, and the future will
exist, it is not true that either do exist, or from a view like the growing block model, according to
which the past and present do exist but the future does not exist.
1
For example, the view is defended in some form by Putnam (1967), Mellor (1998) and Sider (2003).
1
While this way of glossing the ontic component is a useful first start and gives something of the
flavour of eternalism as it is most commonly defended, we can do better. Notice that the claim that the
past, present and future all (unrestrictedly) exist is, at least on the face of it, consistent with it being
true, at every time within a world, that the past, present and future exist, with it nevertheless being the
case that the past, present and future that exist are different at each such time. That is, the ontic
component so construed is compatible with the idea that the ontology of the world changes.
Indeed, it is compatible with two senses in which the ontology might change. First, it is
compatible with ontic component (1) that the world is characterised by a change in how things exist.
More specifically, it is compatible with the ontic component so construed that the ontology of the
world changes in virtue of the fact that times (and the events/objects located at those times) are
constantly gaining or losing the intrinsic temporal properties of pastness, presentness and futurity. So,
for instance, ontic component (1) is compatible with worlds in which there is a “moving spotlight” of
presentness, which shines on progressive instants. At each moment, the instant which is “lit up” has
the property of presentness, and each moment before it has the property of pastness, and each moment
after it has the property of futurity.
Ontic component (1) is also compatible with a more dramatic sense in which the ontology of
the world might change, namely that parts of reality might come into- or go out of- existence. For
instance, in a shrinking tree world the past, present and multiple futures all exist, and future branches
‘drop off’ as the present moves, leaving only a single, unique history (McCall 1996). Thus, on this
view the sum total of reality gradually decreases. Alternatively, in a growing block world the sum
total of reality is constantly increasing; as the present moves new ‘slices’ are continually added to the
‘growing salami’ of the world2. On both views then the totality of events and objects that exist when
one location is the present is different to the totality of events and objects that exist when a different
location is the present. That being said, however, of this pair of views only the shrinking tree view is
consistent with ontic component (1), since growing block worlds are ones in which the future does not
exist and (1) states that the past, present and future all (unrestrictedly) exist.
Now, we take it to be a key feature of eternalism that the ontology of the world does not
change in the more dramatic sense: all and only the times that unrestrictedly exist, exist simpliciter
(we leave it open as to whether the ontology changes in the weaker sense, more on this below). Or to
put the point another way, relative to any time within a world, the very same set of times exists
(unrestrictedly). So if the dinosaurs were to provide a list of all the times that (unrestrictedly) exist,
and if some future robot were given the same task, the two lists would be identical. That is the sense
in which the ontology of the world does not change: the totality of times that compose the world is
always the same; from an atemporal god’s eye perspective, the world does not change. This suggests
the following modification to the ontic component:
2
The growing block view is defended by Forrest (2004), Tooley (1997) and Button (2006).
2
Ontic Component (2): The past, present and future exist and the set of facts regarding which
temporal locations exist is unchanging.
Although this modification to the ontic component is close to what we want, we can still do better.
The difficulty with both versions considered thus far is that both make reference to the past, present
and future. On the face of it, this seems to suggest that there is some objective ontological distinction
to be drawn between these three categories. For example, ‘being past’ might turn out to involve the
instantiation of some special property of pastness. Although, there is scope to include an objective
distinction between past, present and future within an eternalist model of time (as we shall see later
on), most eternalists do not think that past, present and future are objective features of reality. As
such, it would be better if we could phrase the ontic component of eternalism in a manner that is
neutral between these two ways of developing eternalism. Hence, in order to avoid confusion it is
useful to eradicate any mention of the past, present or future from our statement of the ontic
component. Thus we suggest replacing Ontic Component (1) and (2) with:
Ontic Component (3): w is an eternalist world only if, quantifying unrestrictedly, for every
time t that did, does, or will exist, t does (tenselessly) exist.
This statement of the ontic component does not yet take into account the idea that the ontology of the
world is static, and, what’s more, it replaces the notions of pastness, presentness and futurity with
mention of what did, does or will exist: that is, with further tensed locutions. Thus we suggest
modifying the ontic component (3) to give us:
Ontic Component (4)3: w is an eternalist world only if there is a set S of times t 1..tn such that
at every t, all and only the members of S unrestrictedly exist.
Ontic Component (4) looks pretty good4: it captures the core idea behind a basic form of eternalism.
However, most eternalists endorse a further modification to the Ontic Component in light of recent
3
4
Notice that it is consistent with this way of explicating the ontic component that a world in which there is
just one time meets the criterion. That need not mean that such a world therefore counts as an eternalist
world. This is because one of the other components of eternalism, the B-series component, places constraints
on whether or not a world is an eternalist world and that component might rule out there being an eternalist
world in which only a single time exists (more on this below).
One might argue that Ontic Component (4) fails to draw a meaningful distinction between the ontology of an
eternalist world and the ontology of a presentist world. According to the presentist, there exists
(unrestrictedly) a single time t. Moreover, at t, there is a set S of times t1...tn such that at every t, all and only
the members of S unrestrictedly exist. It is just that, for the presentist, S is a singleton set, since there is only
one time in existence: the present time. However, we take it that presentist worlds have been ruled out in
virtue of the fact that the more drastic form of ontological change, according to which things come into- or
3
developments in physics. As stated, the ontic component makes reference to times. Stating the ontic
component in this manner, however, is not sufficiently sensitive to the idea that talk of ‘times’ may be
misleading in worlds in which time and space are fundamentally interwoven as, for example, in
worlds like our own in which it makes better sense to talk of space-time points than of times.
In order to see the idea here, it is useful to provide a brief overview of the structure of spacetime. According to the special theory of relativity, the spatial and temporal distances between events
depend upon one’s inertial frame of reference (roughly, a coordinate system in constant motion). It is
this frame-relativity of distance, combined both with the claim that the speed of light is a constant and
with the claim that the laws of nature are the same in every frame of reference, that gives rise to one
of the core features of relativistic mechanics: the relativity of simultaneity. Hermann Minkowski
noticed, however, that although there is no frame-invariance with regard to spatial distance or
temporal distance, there is a frame-invariant spatio-temporal distance to be had.
Minkwoski (1908) went on to develop this frame-invariant distance between events into a
four-dimensional space-time manifold within which time is treated as a dimension in the manifold
along with the other three spatial dimensions. To represent the connectedness of space-time points in
Minkowski space-time we think of every space-time point as both the point of convergence and the
point of emission of a spherical wave-front of light: what we call the backwards and forwards lightcones.
In any Minkowski manifold, M, there is a value, S, for the distance between any two points.
There are three classes of values for S within M: positive, negative and null. Each of the three
possible values for the spatio-temporal distance, S, between space-time points is represented by a
different feature of the light-cone model. First, space-time points that are at null spatio-temporal
distance from one another are represented by the surface of both the backwards and forwards lightcones. These space-time points are said to be light-like separated because for any space-time points x
and y that are at null spatiotemporal distance from one another, x can only be reached from y by a
signal travelling at the speed of light.
Space-time points that are separated by a positive spatio-temporal distance are said to be
'time-like' separated. Points that are time-like separated fall on the interior of each other’s light-cones.
Time-like separations are such that for any space-time points x and y that are time-like separated, x
can be reached from y by a signal travelling slower than the speed of light.
Space-time points that are separated by a negative spatiotemporal distance are said to be
‘space-like’ separated and are represented by the area outside both the forward and backward lightcones. This area is a space-time point’s absolute elsewhere and is such that for any space-like
go out of existence as the present moves, has been ruled out. This is because presentist worlds are
necessarily worlds in which there is drastic ontic change of this sort. For further discussion of the distinction
between presentism and eternalism see Sider (2006).
4
separated points x and y, x can only be reached from y by a signal travelling faster than the speed of
light. Thus, since no causal influence can travel faster than the speed of light, there can be no
causation between space-like separated space-time points.
Importantly, space-time points that fall on either the interior or the surface of x's light-cone
can be further divided into two categories. On the one hand, all observers in all inertial frames of
reference will agree that events that occur at space-time points that are on x's backward light-cone
occur before x. As such, the class of events that are on x's backward light-cone are in x's absolute
past. On the other hand, all observers in all inertial frames will also agree that the events on x's
forward light-cone occur after x. As a consequence, we call x's forward light-cone x's absolute future.
Now, strictly speaking, an eternalist world need not be a world in which space and time are
combined to form a four-dimensional spatio-temporal manifold. However, a world in which there is a
Minkowskian space-time ought to count as an eternalist world and, one might argue, the manner in
which we have stated Ontic Component (4) does not do justice to this fact. In so far as this is a
concern, we suggest the following modification of the ontic component:
Ontic Component (5): w is an eternalist world only if (A) w is such that for every time t that
exists in w, there is a set S of times t1..tn such that at every t, all and only the members of S
unrestrictedly exist or (B) w is such that for every space-time point p that exists in w, there is a
set S* of space-time points p1...pn such that at every p, all and only the members of S*
unrestrictedly exist.5
Ontic Component (5) explicitly accounts for eternalist worlds in which there are space-time points
and worlds in which there are times and thus it captures the ontological foundations of eternalism, at
least as that component is generally understood. It is important to note, however, one potential
deficiency of Ontic Component (5). In framing Ontic Component (5) by appealing to times or spacetime points one must presuppose that eternalist worlds are worlds in which there is temporality.
However, this assumption may be false: it is an open question whether a world in which there is no
time nevertheless counts as an eternalist world. Thus, although Ontic Component (5) captures the
ontological foundations of eternalism as it is most commonly developed, we will have reason to
revisit the ontic component of eternalism in Section Four, where we discuss the possibility of timeless
eternalist worlds.
Before proceeding to a discussion of the B-series component of eternalism, there is one final
issue that it is worth commenting on. An eternalist world is often called a block universe. There are,
5
It is worth noting that none of these construals of the ontic component rule out presentist worlds as being ones
that meet this particular component. So long as the set of times S in question can be a singleton set of a single
time, then a presentist world meets ontic component (5). Since this component is only part of a necessary
condition for counting as an eternalist world and not a sufficient condition this is as yet not a problem. We
revisit this issue later.
5
we think, two possible things that one might mean by calling an eternalist universe a block universe.
The first is that eternalist worlds are worlds in which there is a four-dimensional spatio-temporal
manifold. If this is what is meant, however, then it would be a mistake to suppose that “eternalist
world” and “block universe” are interchangeable phrases. This is because there are perfectly good
eternalist worlds in which there is no such thing as space-time (such worlds are, perhaps, worlds in
which relativistic mechanics is false).
The second thing one might mean by calling eternalist worlds block universes concerns the
geometry of the universe. Roughly, on this view a block universe is one in which space-time does not
have any intrinsic, mean curvature. Rather, space-time is, roughly, Euclidean. Again, we think it
would be a mistake to identify eternalist worlds with block universes in this sense. This is because
there is a range of solutions to Einstein’s field equations (roughly, the equations that describe the
gravity of a space-time based on the distribution of matter and energy) in which the space-time has
some curvature, imbuing it with a different geometric and topological structure to a Euclidean spacetime. These worlds, we think, are perfectly good candidates to be eternalist worlds: the geometric
features of the universe are not essential to the characterisation of eternalism.
2.
The B-series: Sophisticated vs. Basic
The second component to a definition of eternalism is some sort of commitment to the B-series. The
B-series is a series of times6 ordered by the relations of earlier than, later than and simultaneous with
such that these relations are not subject to change (McTaggart 1908, p. 458). That is, if for any two
times t and t*, if t is earlier than t*, t cannot ‘change’ its position in the B-series in such a manner that
it becomes later than t*, say. At a first pass then, let us say that:
B-series component (1): w is an eternalist world only if w is a world in which there is a Bseries.
The basic form of the B-series employed in eternalism is usually thought to have at least three further
features: First, the B-series is thought to be a linear ordering. Second, the B-series is usually thought
to have an in-built asymmetry: the series runs from the beginning of the universe to the end of the
6
One might also construe the B-series as an ordering of objects/events rather than times. Whether or not one
chooses to construe the B-series in this manner depends, we think, on whether on is a substantivalist about
time or whether one is a relationist. If one is a substantivalist one believes that time exists independently of
the objects/events in time. If one is a relationist, however, then one thinks that all there is to time is the
relations between objects and events. Clearly, if one is a relationist then a B-series ordering of objects/events
is to be preferred whilst, if one is a substantivalist, one will prefer to construe the B-series as an ordering
over times. Unfortunately, the debate between substantivalism and relationism is well beyond the scope of
this paper. As such, since most eternalists seem not to be relationists, we will construe the B-series as an
ordering of times rather than objects/events. However, everything we have to say about eternalism can, if
one wishes, be recast using objects/events rather than times without loss.
6
universe and not vice versa (the clearest manifestation of this asymmetry is in the causal ordering of
events across the B-series: for any x and y such that x is earlier than y, x can cause y but not vice
versa). Third, the B-series is thought to be unique: there is only one B-series in a basic eternalist
world.
Let us call the combination of these three B-series features the simple B-series. Further, let us
call any world that satisfies the ontic component of eternalism and which is one in which there is a
simple B-series, a simple eternalist world. Finally, let us call the view simple eternalism, which takes
the simple B-series to be an essential feature of eternalism and thus characterises the B-series
component of eternalism as follows.
B-series component (2): w is an eternalist world only if w is a world in which there is a simple
B-series ordering of events.
For the simple eternalist, all eternalist worlds are simple eternalist worlds. For those eternalists for
whom the simple B-series is not an essential feature of eternalism, the simple eternalist worlds are a
proper subset of the eternalist worlds.
As we relax which of these three features of the simple B-series we take to be essential to
eternalism we get progressively more sophisticated version of eternalism. Consider, first, uniqueness.
The B-series is unique in w if there is just one B-series in w. However, if one considers the actual
world with respect to this further constraint, one already finds reason to be sceptical. Both the special
and general theories of relativity entail the existence of multiple B-series orderings, each of which is
equally correct. One can see why by focusing on the special theory of relativity (henceforth STR). As
already noted, according to STR the spatial and temporal distances between events are not absolute.
Rather, spatial and temporal distances turn out to be relative to an inertial frame of reference. So, for
example, suppose that we have two events: E and E*. Suppose further that there are two observers O 1
and O2 such that O1 and O2 are in constant motion relative to one another. According to STR, if E and
E* are simultaneous for O1 then they will be non-simultaneous for O2: the B-series ordering of the
two events differs between the two frames of reference. But here’s the rub: both frames of reference
(O1’s vs. O2’s) are on a par according to STR (and general relativity). In particular, the laws of nature
are the same in both frames of reference and the speed of light is constant across reference frames.
Thus, there is no physical basis upon which to mark off one B-series ordering of events as the
‘correct’ ordering. Thus, there are multiple, equally good B-series orderings available.
Since eternalists usually take the actual world to be an example of an eternalist world par
excellence and since the actual world is a world in which STR holds, this either provides us with good
reason to reject the uniqueness component as being part of the characterisation of the simple B-series,
7
or it gives us reason to reject the idea that the simple B-series is an essential component of
eternalism7.
Consider now linearity. It is usually assumed that an eternalist world is one in which there is a
linear topology. But again, it is unclear that one ought to take the linearity component of the simple Bseries to be essential to a characterisation of eternalism, for there are many other possible topologies.
These include closed topologies in which there are closed time-like curves: curves in space-time such
that an object travelling forward in its personal time will end up back in time meeting its earlier self.
A Gödel world is a world like this: a constantly rotating world in which the very same, numerically
identical event is both the ‘start’ and ‘finish’ of a particular temporal sequence8. For such a sequence
of events, there is no consistent linear ordering. That is because, we suppose, a linear ordering is one
in which, minimally, the irreflexivity of the earlier-than and later-than relations is preserved.
There are also topologies in which time fails to be unified. Unification is the temporal
analogue of connectedness in space. A region of space is connected just in case for any two spatial
locations, there is a spatial relationship between those locations, or, as we might say, in case there is a
continuous spatial path joining the two. Two temporal locations are unified just in case there is some
temporal relation between the two, that is, just in case relative to some frame of reference, any
temporal locations t1 and t2, t1 is either earlier than t2, or later than t2, or t1 and t2 are simultaneous. We
can also define a disunified space-time. A disunified space-time is a space-time in which the manifold
itself splits in two, producing two ‘branches’ of space-time such that there is no direct space-like,
time-like or light-like connections between points on the two branches (of course, there are indirect
connections between the points, it is not as though the two branches constitute distinct space-times).
Now, it is not clear to us why a disunifed world ought to fail to count as an eternalist world
or, indeed, why a Godel world should fail to be an eternalist world just because the B-series is nonlinear in such a world. So in both cases, we think, there is pressure to reject the simple B-series as an
essential feature of any eternalist world.
7
Indeed, one might go further and wonder how seriously we should be taking the B-series. In particular, one
might think that due to the perspectival nature of the B-series (i.e. each ordering is relativised to a particular
perspective: an inertial frame of reference) the metaphysical robustness of the B-series has been thinned
down. To see the idea here consider the following: suppose that we have before us two moral views. On the
one hand, there is a view according to which there is just one set of moral facts. On the other hand, we have
a view according to which there are multiple (and incommensurate) sets of moral facts, each of which is
equally good (perhaps the facts are relativised to cultures, nations or individuals). Although in both cases we
have moral facts, one might be inclined to think that there is a difference in strength between the two views.
Accepting that there are many, equally good moral views is, one might think, to take moral facts ‘less
seriously’ than if one were to accept that there is just one moral view. This applies also to the B-theory:
having many, equally good B-series is to take the ordering at issue ‘less seriously’. Even if this is correct,
however, we have little choice: the physics demands that we give up the basic form of the B-theory and
endorse a more sophisticated and, perhaps, metaphysically weaker form of that view according to which
there are many, equally good B-series in the world.
8
Godel worlds are space-times that are characterisable by Kurt Godel’s solution to Einstein’s field equations.
For further discussion of Godel worlds see Bourne (2006).
8
Finally, consider asymmetry. That the B-series is asymmetric is also, we think, inessential to a
characterisation of eternalism. For example, one might defend a view according to which there is no
sense in which the B-series runs from the start of the universe to the end of the universe. Rather, the
B-series simply ties the two ends of the universe together with no regard for directionality9 (a view
along these lines is developed and defended by Huw Price (1997)). Alternatively, one might accept
that there is asymmetry but that it runs in the opposite direction: moving from the end of the universe
back towards the beginning of the universe. Finally, one might think that we live in a Gold universe
(so-called after Thomas Gold the cosmologist who proposed the model in the 1960s). Strictly
speaking a Gold universe is one in which the universe begins with a big bang, expands with
increasing entropy for some time until it starts to contract and entropy then decreases until it reaches a
low entropy state and ultimately a big crunch. If one thinks of the direction of time as being grounded
in some way in the thermodynamic arrow, then a Gold universe is one in which the B-series runs in
one direction until time t at which time it reverses direction. More accurately, one can think of a Gold
universe as a universe in which there are two B-series running in opposite directions from the two
ends of the universe, meeting somewhere in the middle.
There is a cluster of different eternalist views on offer depending on which combination of
these features of the simple B-series one rejects as being essential to eternalism. We will not consider
each of these here. Rather, we distinguish only two, which we call standard sophisticated eternalism
and weak sophisticated eternalism respectively. Standard sophisticated eternalists reject as essential to
eternalism just the uniqueness component of the simple B-series. Thus they hold that there are
eternalist worlds in which there are many equally good B-series orderings, one for each inertial frame
of reference. But there are no eternalist worlds in which there are B-series orderings that are nonlinear or symmetrical. We can thus characterise the B-series component of eternalism according to the
standard sophisticated eternalist as follows:
B-series component (3): w is an eternalist world only if there is a B-series ordering of times in
w, and the ordering of those times is both linear and asymmetric.
We call this view standard sophisticated eternalism because, as a sociological matter of fact, we think
this is the view that most eternalists hold. This is in contrast to weak sophisticated eternalists who
reject the uniqueness, linearity and asymmetry constraints and thus are committed only to the
following B-series component of eternalism:
9
One might think that the irreflexivity of the earlier-than and later-than relations follows from the failure of
asymmetry. However, this is incorrect. One can have a world in which temporal relations failure to have a
direction at all without those relations needing to be reflexive.
9
B-series component (4): w is an eternalist world only if there is at least one B-series ordering
of events in w.
It is worth noting at this point that nothing we have said so far in characterising eternalism precludes
there being eternalist worlds in which there is, in addition to a B-series, also an A-series. We consider
this issue further in the following section. For now, however, we should point out that again, as a
sociological matter of fact, standard sophisticated eternalists either hold that necessarily, there is no
A-series, or at the very least, that for every eternalist world w, there is no A-series in w.
3.
The B-series: Flow?
In the previous section we drew a distinction between simple eternalism and two versions of
sophisticated eternalism. In this section we draw a further distinction between two different kinds of
eternalism that we will call inclusive and exclusive. So far, we have assumed that eternalists take the
existence of the B-series in a world to be necessary for that world to be an eternalist world. But the
existence of a B-series in a world does not preclude the existence of an A-series in that world. The Aseries is, like the B-series, a temporal ordering. The A-series, however, orders times based on whether
they are past, present or future (rather than earlier than, later than or simultaneous with some other
position) (McTaggart 1908, p. 458). Unlike the B-series, the A-series is subject to change. In
particular, positions in the A-series change from being future to being present to being past. This
change is usually identified with the flow of time.
Let us call the B-series exclusive just in case necessarily, for any world w in which there is a
B-series, there is not also an A-series. Let us call the B-series inclusive just in case it is not the case
that necessarily, for any world w in which there is a B-series, there is not also an A-series. Then we
now have two amended versions of each of the competing B-series components ((1) through (4))
mentioned in the previous section corresponding to whether we take the B-series in question to be
exclusive or inclusive. However, since we are focusing on sophisticated eternalism, the more
important of these would then be:
B-series component (3a): w is an eternalist world only if there is an exclusive B-series
ordering of events in w, such that the ordering of those events is both linear and asymmetric.
B-series component (3b): w is an eternalist world only if there is an inclusive B-series
ordering of events in w, such that the ordering of those events is both linear and asymmetric.
B-series component (4a): w is an eternalist world only if there is an exclusive B-series
ordering of events in w.
B-series component (4b): w is an eternalist world only if there is an inclusive B-series
ordering of events in w.
10
(3a) represents the strongest construal of the B-series component that we will consider and (4b) the
weakest. As we understand matters, most philosophers who self ascribe as eternalists accept
something closer to (3a) than any of the other three ways to construe the B-series component. Notable
exceptions to this are those who reject the exclusivity of the B-series and thus adopt either (3b) or
(4b).
It is important to point out at this junction that there are two very different models of time
compatible with both (3b) and (4b). These models differ over how one conceives of the A-series.
Specifically, they differ over whether or not the A-series itself is taken to be unique10. If the A-series
is unique, then there is a single way of ordering times in terms of whether they are past, present or
future. On such a view the flow of time captured by the A-series is conceived of as a universe-wide
‘wave’. This view accords with the Moving Spotlight or Hybrid view of time11. Or, more carefully,
according to an eternalist who accepts the inclusivity of the B-series, the Moving Spotlight and
Hybrid worlds are to be classified as eternalist worlds despite their having an A-series. Worlds that
lack an A-series (if there are any) and meet the B-series component and the ontic component will, of
course, also count as eternalist worlds.
If, however, the A-series is non-unique, then there is no single way of ordering times in terms
of whether they are past, present or future. Rather, there are many, equally good orderings of this
kind. The flow of time, on this view is more like a system of capillaries, with numerous ‘channels’
down which there can be a moving present12. Either of these ways of thinking of the A-series is
consistent with both (3b) and (4b). Even in a world in which the B-series is non-linear, asymmetric
and non-unique, (as if (4b) holds) one can suppose there is a unique A-series in virtue of there being a
unique metaphysically privileged frame of reference that determines the A-series, though no unique
physical frame that determines a unique B-series13.
Whether or not an eternalist takes the B-series component of eternalism to require an
exclusive B-series or merely an inclusive B-series depends on a couple of factors. First, clearly, it
depends on whether she thinks that there are any possible worlds in which there is an A-series, and
thus any worlds in which there is both an A-series and a B-series. If one holds that the A-series is
10
11
12
13
As well as uniqueness, one might also think of the A-series in terms of linearity. Specifically, just as one
might relax the assumption of uniqueness, one might also construe the A-series as non-linear (think of a
moving present that travels in a loop). Note, however, that the A-series is usually thought to be asymmetrical
and, furthermore, that this seems to be an essential feature of the A-series. If the A-series is not
asymmetrical then the flow of time has no direction. However, having a particular directionality would seem
to be essential to temporal flow.
The moving spotlight theory of time is discussed in McKinnon (1999), Skow (2011), and McTaggart
(1908).
The central question for such a view is whether or not the uniqueness of the A-series really is up for grabs.
One reason for thinking that it is not is that there is supposed to be something special about the objective
present. However, if there are multiple objective presents to be had then prima facie it may be hard to do
justice to the idea that the present is metaphysically special.
A view along these lines is defended by Craig Bourne (2006, p. 183).
11
internally inconsistent and hence impossible, then one ought to reject inclusivity in favour of
exclusivity. If one countenances the existence of worlds with an A-series and a B-series, then the
question remains as to whether those worlds ought to count as eternalist worlds or not and thus
whether or not to embrace inclusivity in the B-series. If one thinks it is essential to a characterisation
of eternalism that eternalist worlds lack an A-series, then one will reject any worlds with an inclusive
B-series as being eternalist worlds even while admitting that there are such worlds. If, however, one
thinks that such worlds are eternalist worlds, then one will embrace the somewhat weaker inclusive
characterisation of the B-series component of eternalism.
Of course, it is consistent with the inclusivity of the B-series that every world with a B-series
also has an A-series (rather than merely some of them). Indeed, if one were an eternalist who
supposed that the A-series is essential to the existence of temporality, then one would either embrace
the inclusivity of the B-series (and take it that every eternalist world has both a B- and an A-series) as
a component of eternalism or adopt an error theory about temporality (if one thinks that the A-series is
internally inconsistent a la McTaggart (1908, pp. 467-70)). This, however, is unlikely to be true of
any eternalist we know, since as a matter of fact, most eternalists think that the A-series is not
essential to the existence of time. Thus, most eternalists think that time does not (and in the case of
many eternalists, necessarily does not) flow in any meaningful sense.
Note, though, that the spectre of being error theoretic about temporality raises some further
interesting questions about the role of the B-series in a characterisation of eternalism. The various Bseries components that we have considered above yield characterisations of eternalism according to
which some kind of B-series is essential to a world’s counting as an eternalist world. In the following
section we consider whether any satisfactory characterisation of eternalism must include a B-series
component.
4.
The B-series: Limiting Cases
Consider a world in which there is a single three-dimensional slice: a three-dimensional ‘snapshot’ of
the world at a time. Call such a world a W1-type world. Is a W1-type world an eternalist world? On the
face of it, a W1-type world is not a world in which there is a B-series. This is because in order for
there to be a B-series in a world there must be some irreflexive relations of earlier-than and later-than
in that world (even if all of these relations are not irreflexive, as in a Gödel world). However, prima
facie there are no such relations in a W1-type world. For there is only a single time in such a world,
and all events/objects in that world exist at that one time.
Here’s a very quick argument for the view that a W1-type world is an eternalist world.
Suppose, for a moment, that the actual world, W@, is a simple eternalist world: a world in which there
is a single B-series. Such a world can be conceptualised as a ‘stack’ of three-dimensional ‘slices’.
Now, imagine that we produce a second world, Wa, by taking just one of the slices away. Is Wa an
12
eternalist world? On the face of it, the answer appears to be ‘yes’. This is because, intuitively, the loss
of a single temporal slice cannot make the difference between whether a world is an eternalist world
or not. But suppose that we produce another world, Wb, by taking a slice from Wa. Is that world an
eternalist world? Again, the answer seems to be ‘yes’. The trouble is that we can keep taking slices
away to produce new worlds. But if the presence/absence of a single slice can never make the
difference between whether or not a world is an eternalist world, then by following the same soritesstyle reasoning, we can construct a simple eternalist world by taking a world in which there are only
two three-dimensional slices and removing one of those slices.
You don’t need to take sorites style arguments seriously to wonder whether, if a two-slice
world is an eternalist world, then a one-slice world is also an eternalist world. Whether one finds it
plausible that removing a single slice will not make the difference between a world being an eternalist
world and not being an eternalist world ultimately depends on whether one thinks that necessarily,
any eternalist world has a B-series. For the only relevant difference being made by the removal of a
single three-dimensional slice from a world in which there are two such slices is that removing a slice
strips that world of any relations of earlier-than and later than. Hence, if one’s intuition is that the
conclusion of the sorites style argument is false, this reveals that one considers the B-series to be
essential to a characterisation of eternalism. Mutatis mutandis if one thinks that the argument goes
through.
A similar line of reasoning can be applied to sophisticated eternalism. In that case, however,
rather than thinking in terms of a single three-dimensional slice, it is more useful to think in terms of
space-time points. The usual way of thinking about a sophisticated eternalist world is as a fourdimensional manifold, the positions in which have both a spatial and temporal component. Such a
manifold is, essentially, a Minkowski space-time. In Minkowski space-time the issue of what counts
as a limiting case of an eternalist world may not revolve around consideration of a world with just one
three-dimensional slice. For it only makes sense to talk of a three-dimensional ‘slice’ from within a
frame of reference. This is because there is no frame-invariant way to ‘slice up’ the manifold into
three-dimensional slices in this manner. Rather, there are equally many good ways to carry out such a
slicing (this is, really, just another way of saying that there are equally many good B-series orderings
in Minkowski space-time).
However, this has the following consequence. Consider a one-slice, sophisticated eternalist
world. In such a world, there is a frame of reference, R1, in which all of the events that exist in the
slice are simultaneous. There are, however, many frames of reference in which the events in question
are not all simultaneous with one another. For example, there is a frame of reference R 2 in which
some of the events stand in earlier-than or later-than relations with one another. If this is correct,
however, then such a world is not, strictly speaking, a world without a B-series. At best, it is a slightly
impoverished sophisticated eternalist world in which there is one less B-series ordering than there
would usually be.
13
As such, the interesting case will be to consider whether a world with a single space-time
point counts as an eternalist world. We arrive at such a world by progressively removing space-time
points from the manifold of a sophisticated eternalist world. Here again, a world with a single spacetime point is not a world in which there is a B-series. For consider, a single space-time point
decomposes into a spatial and a temporal component. Thus, in such world at best all that exists is a
single spatial location and a single temporal location. But a single temporal location is not sufficient
for the existence of any B-relations of earlier-than and later-than (regardless of one’s frame of
reference). This is because these relations can only ever hold between distinct temporal locations.
Since worlds that are paradigmatically considered to be eternalist worlds do have a B-series, it
is unsurprising that intuitions might differ with respect to these worlds that are either limiting cases of
eternalist worlds (if one thinks the B-series is non-essential to eternalism and thus rejects the various
B-series components altogether) or worlds that are not eternalist worlds at all (if one thinks the Bseries is essential to eternalism and thus accepts some version of the B-series component). That
people turn out to deploy somewhat different concepts, all under the name “eternalism” is certainly
worth noting. It is not obvious which of these concepts is the best or most natural one. One thing that
seems right to us, however, is that the weaker one’s conception of the kind of B-series that is essential
to eternalism, the more unstable one’s position and the more likely to collapse into a position
according to which there is no essential B-series component to eternalism. Once one allows that the
only kind of B-series that is essential to eternalism is one that does not require uniqueness or linearity
or asymmetry or a particular kind of topology, then it is more difficult to resist the final move
according to which the B-series is not essential to eternalism at all. For the sense of “temporality” and
“tense” that one can extract from worlds with such B-series are already very unlike our ordinary
everyday sense of these concepts. On the other hand, it is very easy to see how to resist such a move if
one embraces something like B-series component (2) or B-series component (3a). For the difference
between a world with no B-series such as the one slice world or the one space-time point world, and a
world that is characterised by (2) or (3a) is radically different in kind.
So it may be that eternalists are faced with something of a dilemma. On the one hand, they
can accept a very “meaty” conception of the nature of the B-series, and take that series to be essential
to eternalism. But the worry about this is that it is very doubtful that the actual world would then turn
out to be eternalist given what we now know about its physics. On the other hand, the eternalist can
accept a more liberal conception of the B-series. But once she does that, it becomes more difficult to
see a way to draw a principled line that rules in as eternalist worlds only those worlds with such a Bseries, and rules out worlds that lack any such series.
Let us call the view that the B-series (in any form at all) is inessential to eternalism an
atemporal conception of eternalism. If one is an atemporal eternalist, then one is faced with some
extra work. For the atemporal eternalist rejects all of the B-series components. Thus they define
14
eternalism entirely in terms of the ontic component. But the ontic components we have offered will
simply not do the job. Consider the most sophisticated of these, ontic component (5):
Ontic Component (5): w is an eternalist world only if (A) w is such that for every time t that
exists in w, there is a set S of times t1..tn such that at every t, all and only the members of S
unrestrictedly exist or (B) w is such that for every space-time point p that exists in w, there is a
set S* of space-time points p1...pn such that at every p, all and only the members of S*
unrestrictedly exist.
Intuitively, in an atemporal eternalist world clause (A) may not obtain, since there are might be no
times in such a world. But likewise, clause (B) will also fail since if there is no temporality in a world,
it is hard to see how there could be space-time points. As such, what we need is a form of the ontic
component that leaves open the possibility that the universe might consist of some space of points
such that there are no temporal or spatio-temporal metric relations of any kind between those points
but nevertheless, in typical eternalist style, all of the points exist unrestrictedly. Since the ontic
component is now the only component that defines eternalism, we are looking for a definition that
gives us both necessary and sufficient conditions for a world’s being eternalist. As a first pass we
might modify Ontic Component (5) along the following lines:
Ontic Component (6): w is an eternalist world if only if (A) w is such that for every time t that
exists in w, there is a set S of times t1..tn such that at every t, all and only the members of S
unrestrictedly exist or (B) w is such that for every space-time point p that did, does or will exist
in w, there is a set S* of space-time points p1...pn such that at every p, all and only the members
of S* unrestrictedly exist or (C) w is composed of a space C such that (i) at every point p* in C
that exists there is a set S** of points p*1…p*n such that at every p*, all and only the members
of S** unrestrictedly exists, and (ii) there are no temporal or spatio-temporal metric relations
between points in C and (iii) the points of C have spatial dimensions only.
It is useful to consider some examples of atemporal worlds that meet clause (C). Suppose that
the space C is a configuration space: roughly, a space that describes all of the physically possible
configurations of some physical system P. More specifically, suppose that C is a configuration space
for all of the particles in the universe. The points in C will thereby be three-dimensional arrangements
of those particles. Such arrangements accord with what we might think of as times. However, in a
configuration space there need not be any temporal or spatio-temporal metric relations between the
points in that space. So we need not think of the points in C as times (or, indeed, space-time points),
even though that is what they look like. A configuration space of this kind satisfies (C), and as such
there are two kinds of world that will count as eternalist given ontic component (6). First, an eternalist
15
world w might be such that (a) all of the physical possibilities are realised and (b) there are no metric
relations between points in C (although the points themselves are constituted by spatial metric
relations between the particles being configured) and (c) the points in C have spatial dimensions only.
Second, an eternalist world w might be such that (a) all of the physical possibilities are realised and
(b) although there are no temporal or spatio-temporal relations between points in C, there are spatial
metric relations between points in C and (c) the points in C have spatial dimensions only.
Call the first kind of world an atemporal Barbourian eternalist world (Barbourian because
this is very close to a description of what Barbour thinks our world is like, more on this below). Call
the second kind of world an atemporal hyper-eternalist world. The difference between these two
worlds concerns the metric relations between points in the configuration space. In the Barbourian
world the points in the space ‘float free’ of one another; they are spatially and temporally isolated. In
the hyper-eternalist world, by contrast, the points are connected via a fourth purely spatial dimension
(hence, such a world is a world in which there could be four-dimensional objects, such as Klein
bottles or Tesseracts). Note that there are a large number of hyper-eternalist worlds compatible with
clause (C). This is because there is no limit to the spatial dimensions of the points in a configuration
space. Thus, if we start with three-dimensional points, then a hyper-eternalist world will be fourdimensional. However, if we start with four-dimensional points (where all of the dimensions are
spatial as per (iii) in clause (C)) then a hyper-eternalist world will be five-dimensional. More
generally we can say that for any configuration space in which the points are n-dimensional, a hypereternalist world will have n+1 spatial dimensions.
Which of these versions of atemporal eternalism is to be preferred depends upon how
uncomfortable one is with the idea that the points in C bear no metric relations to one another. For if
there are no metric relations between the points, then they are spatially and temporally isolated, even
though they exist. As such, one might be inclined to think that Barbourian eternalism is really a form
of concrete modal realism: the view according to which there are spatially and temporally isolated
universes. This is because, given the lack of metric relations between points in C, those points might
seem to be distinct three-dimensional worlds.14 Thus, in so far as one is worried about atemporal
worlds collapsing into modal realism one should prefer the hyper-eternalist alternative.15
At this point one might worry about using configuration spaces to define eternalism. This is
because eternalist worlds that meet clause (C) will be worlds in which all of the physical possibilities
are realised. Or, rather, such worlds are worlds in which all of the possible n-dimensional
configurations of the world are realised (in some cases: in n+1 dimensions). But an eternalist who
14
Though of course, this is a concrete modal realism in which only the physically possible worlds exist, at least
in that particular configuration space.
15
One might think that the isolation of points in an atemporal eternalist world is simply an artefact of using
configuration spaces to meet clause (C) and that if some other space is used, then the points in the relevant
space will be less isolated. However, this is mistaken: the isolation is due to the fact that there are no metric
relations between points in atemporal eternalist worlds and not due to any intrinsic feature of the
configuration space itself.
16
rejects the existence of time need not be committed to the view that all of the physical possibilities are
realised.
In response, one might try to find some other space to define clause (C). The trouble with this,
however, is that the spaces to which one might appeal are, as with the configuration space, spaces of
possibility. For example, one might appeal to an n-dimensional Hilbert space. However, Hilbert
spaces, like configuration spaces, are possibility spaces. An alternative option for responding to this
worry then might be to define an eternalist world as some sub-set of points in the relevant space
(rather than the set of all points). This suggests the following modification to Ontic Component (6):
Ontic Component (7): w is an eternalist world if only if (A) w is such that for every time t that
exists in w, there is a set S of times t1..tn such that at every t, all and only the members of S
unrestrictedly exist or (B) w is such that for every space-time point p that did, does or will exist
in w, there is a set S* of space-time points p1...pn such that at every p, all and only the members
of S* unrestrictedly exist or (C) w is composed of a set of points S` in some space C such that
(i) at every existing point p* in S` there is a set S** of points p*1…p*n such that at every p*, all
and only the members of S** unrestrictedly exists and (ii) there are no temporal or spatiotemporal metric relations between points in S` and (iii) the points of C have spatial dimensions
only.
Barbourian eternalist worlds, then, are worlds in which the points in S` bear no metric relations to one
another. Hyper-eternalist worlds, by contrast, are worlds in which the points in S` bear spatial metric
relations to one another only. The upshot of defining atemporal eternalist worlds in this manner is that
S` need not contain all of the points in C. So the atemporal eternalist need not be committed to the
view that all of the physical possibilities are realised in any atemporal eternalist world.
5.
Fundamentality
In this section we consider a further, interesting question about how the eternalist should construe the
relationship between the B-series and eternalist worlds. This is the question of the fundamentality of
the B-series. Here there seem to be two options open: the eternalist might suppose that the B-series is
fundamental, or that the B-series is emergent (or supervenient, or derivative). Notice that even an
eternalist who rejects the contention that the B-series is essential to eternalism, will likely hold that
some eternalist worlds have a B-series, and thus ought to be interested in the nature of that B-series.
Exactly what it takes for some x to be fundamental is in itself debatable.16 It might be that x is
fundamental iff x is its own supervenience base. Or it might be that x is fundamental iff the theory
16
See Schaffer
(2009) for a detailed discussion of these issues.
17
that posits x cannot be reduced to any more fundamental theory. Or it might be that x is fundamental
just in case there is nothing that x depends upon. One way we might try to determine what is
fundamental is by looking at the objects posited by our best physical sciences, and in particular, by
looking at the objects posited by our fundamental theories. One might be tempted to suppose that if
our most fundamental theories posit the existence of an entity E, then this is good reason to suppose
that E is fundamental in at least one of these senses, and that if they fail to posit E, that is reason to
suppose either that E does not exist, or that E is derivative on some more fundamental entities.
In looking to the question of whether the B-series is fundamental or derivative, there are two
related questions. Since the B-series is a temporal series, the first is the question of whether time itself
is fundamental or derivative. If time itself is derivative, then this answers the question about the
fundamentality or not of the B-series. On the other hand, if time itself is fundamental, this leaves it
open that the B-series might be either fundamental or derivative, since it leaves it open that, for
instance, the A-series might be fundamental and the B-series derivative or that some other temporal
phenomenon might be fundamental and the B-series derivative. We think it implausible that a world
counts as eternalist in which there is a fundamental A-series and a derivative B-series. That leaves it
open, however, whether there are versions of eternalism in which the B-series is fundamental, and
versions in which it is derivative (but not on the A-series).
The usual assumption is that time is in some good sense fundamental. Sometimes this is
expressed as the claim that our world is fundamentally a four-dimensional one, and that although
space and time are interwoven in a complex manner, there is a time-like dimension in the four
dimensions and that this is important. Sometimes this is expressed in the thought that fundamental
physical theory is characterised with the use of a ‘t’ parameter, and therefore that this temporal
parameter must be fundamental. Not everyone agrees with this, however. For instance, Julian Barbour
(1999) has an interpretation of canonical quantum gravity the first part of which involves
reformulating classical general relativity in a Machian manner17. Leaving the details aside, the idea is
that a Machian formulation of general relativity has no variable that represents time. This is not to say
that given such a formulation it makes no sense to speak of temporality. For there is a sense in which
we can recover or perhaps better, reconstruct, a temporal ordering, and hence time, from such a
formulation. It is just that time is not a fundamental component of the theory. While Barbour
represents this view as temporally eliminativist, it is better, we think, to understand his formulation of
general relativity as instead being one in which time is not fundamental, and thus one in which the Bseries is not fundamental, but instead is derivative.
Barbour’s Machian formulation of general relativity, if correct, would give us reason to think
that the B-series is not fundamental in our world. The question then is whether or not such a world is
an eternalist world. Our view of the matter is that if a B-series ordering can, in some good sense, be
17
For further discussion of Barbour’s view see Baron, Evans, and Miller (2010), Healey (2002) and Ismael
(2002).
18
reconstructed from the fundamental posits then there is a B-series ordering in the world and so even
those who suppose that a B-series is essential to a characterisation of eternalism ought to think that
such a world counts as being eternalist. However, we recognise the possibility of a more hard-line
eternalist view, what we will call serious eternalism. Serious eternalism is the view according to
which a B-series ordering is not only essential to any eternalist world, but in addition such an ordering
must be fundamental.
There is an interesting question here as to whether the B-series is fundamental in a more
standard picture of the nature of space-time, let alone Barbour’s Machian general relativity in which
the fundamental posits are points in configuration space. On the face of it, the fundamental posits of
Minkowski space-time are the space-time points along with the relations between them which define
their connectedness. Of course, one can impose a B-series ordering onto the manifold from within an
inertial frame of reference. To do this, one ‘slices’ or foliates the manifold into a series of threedimensional slices, which can be thought of, roughly, as times, and which are related to one another
by earlier-than/later-than relations. But there is no sense in which that B-series ordering is the
‘correct’ ordering: there are many equally good ones. In the current context, we can see this as the
claim that there is no B-series ordering that is fundamental to the universe. Rather, there are many
equally good ones that supervene on the space-time points.
Regardless, if Barbour is correct then worlds will count as having a B-series, but will do so
despite being very unlike our intuitive picture of a world with temporality. It is these sorts of
considerations that, we think, push the eternalist either towards a very strong definition of eternalism,
or push the eternalist towards jettisoning the idea that the B-series is essential to eternalism (though of
course, many eternalist worlds will be ones with a B-series). As we noted earlier however, the space
of eternalist worlds that correspond to a very strong definition of eternalism is very limited indeed:
only those worlds in which there is a unique B-series ordering. But that would seem to rule out even
the actual world as being an eternalist world.
Returning to the issue of temporality, it is worth noting that both Barbour (Barbour 1999,
Barbour 1994a, Barbour 1994b) and Deutsch (1997) each have a theory of quantum gravity that is
timeless in a much stronger sense: the B-series is neither fundamental nor, they claim, derivative. For
example, Barbour claims that there is no unique way our world was in the past, nor will be in the
future: there is just a static configuration space filled with three-dimensional ‘instants’ and there is no
path through that space that can rightly be thought as a history of a world. Thus there is simply no
temporal ordering in Barbour’s interpretation of quantum gravity. Again, we set aside the details of
Barbour and Deutsch’s quite distinct proposals, and merely note that if there are possible worlds that
are as they describe, then there are possible worlds in which there is no B-series, either fundamental
or derivative, and if Barbour and Deutsch are right, those worlds might seem very much like our
world (indeed, our world might be such a world). Once again, whether such worlds count as eternalist
ones will depend on whether one takes the B-series to be essential to eternalism.
19
6.
Presentism and Eternalism: Indispensability
Thus far we have considered the three main components of eternalism: the ontic component, the Bseries component and the fundamentality component. We provided a precise characterisation of the
ontic component, identified a number of different forms of eternalism that stem from being committed
to different B-series components, and we argued that eternalists should not be serious eternalists; that
is, they should not think that the B-series is fundamental, except in the sense that it should not be
derivative on the A-series.
We turn our attention now to the relationship between eternalism and presentism. Although
presentism has been previously characterised in more detail in this volume, for our purposes we will
characterise presentism in the following manner.
Ontic Component of Presentism (1): w is a presentist world only if w is such that for every
time t that exists in w, there is a singleton set S of times t1..tn such that (i) all and only the
members of S unrestrictedly exist and (ii) for every tn, t = tn.
The Ontic Component of Presentism (1) is not quite right, however. This is because, although there is
only a single time that exists, the presentist maintains that there were past times and will be future
times. To reflect this idea, we can modify the ontic component as follows:
Ontic Component of Presentism (2): w is a presentist world only if w is such that for every
time t that exists in w, there is a singleton set S of times t1..tn such that at every t, all and only
the members of S unrestrictedly exist and for every time t there were past times and will be
future times.
In what follows we present an argument against the ontic component of presentism, an argument that,
to our knowledge, has not been considered hitherto. The argument takes the form of an
indispensability argument against this ontic component, now well known from what is often referred
to as the Quine-Putnam indispensability argument for mathematical Platonism18. The thought, in
rough terms, is that the existence of more than one time is indispensable to our best science and so we
ought to be ontologically committed to the times in question. More formally, the indispensability
argument can be stated as follows:
18
For detailed discussion of the Quine-Putnam indispensability argument see Mark Colyvan
(2001).
20
(P1) We ought to have ontological commitment to all and only the entities that are
indispensable to our best scientific theories
(P2) The existence of more than one time is indispensible to our best science
(C) Therefore, we ought to have ontological commitment to more than one time
Here, we take it that some entity E is indispensable to our best science only if there is some scientific
theory T such that (i) T is a part of our best science (ii) E plays a role in T and (iii) T is more
attractive than some theory T* in which E does not play a role.
(P2) seems plausible. A brief glance at the science will reveal both implicit and explicit
quantification over times (for example in the hard sciences like physics but also in the special sciences
such as biology). Thus, the onus is on the presentist, we think, to carry out a Hartry Field (1980) style
‘nominalisation’ of science: she must show that the existence of multiple times is, in fact, dispensable
to our best scientific theories. To do this the presentist must collect together all of the statements that
seem to involve quantification over non-present times. Call these statements the Ss. One then replaces
the Ss with some class of statements that are capable of (i) playing the same role in the science but (ii)
do not involve quantification over non-present times.
Here’s how this might go. Call quantification over non-present times tensed quantification.
Eternalists have an easy account of tensed quantification: tensed quantification is just restricted
quantification. Presentists, however, cannot understand tensed quantification in this manner. Rather,
presentists analyze tensed quantification as quantification within the scope of a non-truth-functional
tense operator. So, for example, consider the claim that there existed dinosaurs in the past. According
to the presentist, when translating this claim into logical form, we nest the entire quantified sentence
within the sentential operator ‘it was the case that’ (W):
(W) xDx
The idea is that sentential operators of this kind are analogous to non-truth-functional modal
operators, in that nesting a quantifier within the relevant operator undercuts its ontological force
(Sider 1999, p. 326). In the current context, then, the thought would be to replace the Ss with a set of
propositions, the S*s, which use truth-functional tense operators instead of restricted quantifiers.
Although we do not rule out the success of a project of this kind, we are pessimistic that one
can carry out this ‘nominalisation’ of a scientific theory T without rendering that theory less attractive
21
overall. This is so for the following reason. It is likely that the science will involve quantification over
multiple times19 such as:
(S1)
There have been two times in the past at which the moon was between the earth and
the sun
According to David Lewis (2004), however, although tense operators are satisfactory for most
purposes, plural tensed quantification poses a distinct challenge for parsing tensed quantification
along presentist lines (see (Brogaard 2007) for a response to Lewis). For example, in order to parse
(P1) the presentist nests the entire quantified sentence within a past-tensed operator:
(S1`)
(W)(xy((Tx & Ty & x ≠ y) & .... )
But this cannot be right, for (S1`) reads: it was the case that (there are two distinct times such that....).
But prima facie this commits the presentist to the view that it was the case that there are two distinct
times. But this flies in the face of presentism: it was never the case that there are two distinct times.
The difficulty is that, as Lewis goes on to argue, any attempt to sophisticate the general presentist
strategy requires either taking on unattractive theoretical or ontological commitments or endorsing an
analysis of (S1) that has an infinitary construction. This poses a difficulty for the project of
‘nominalising’ science by removing apparent reference to multiple times. This is because, to complete
such a project for some scientific theory T, one must render T a less attractive theory overall, by
running afoul of either simplicity or parsimony.
7.
Conclusion
Despite being both a popular view and, we think, a very profitable avenue of enquiry in metaphysics,
we think that more attention could be paid to thinking about to what one is committed when one says
that one is an eternalist. We have outlined a range of views that are, when all is said and done, very
different from one another. At one end we have a very weak conception of eternalism that will include
as eternalist, worlds in which points in configuration space bear no metric relations to one another. At
the other end of the spectrum we have very strong versions of eternalism that demand not only that all
eternalist worlds have a B-series, but in addition that such B-series’ meet some quite stringent criteria.
We have not, by and large, attempted to adjudicate the question of which of these characterisations is
the correct one, (or the most useful one) beyond noting that some of the strongest characterisations
would seem to preclude the actual world from being an eternalist one, and noting that some of the
19
And indeed, may also require cross-temporal relations, which imposes further constraints on any metaphysical
theory of our universe.
22
weakest would allow worlds extremely unlike ours (or at least, extremely unlike the way our world is
taken to be by most of us) to count as eternalist worlds. We hope that offering these characterisations
fosters renewed interest in thinking about to what eternalists want and ought to be committed.
23
References:
Barbour, J. (1999). The End of Time. Oxford: Oxford University Press.
Barbour, J. (1994a) 'The Timelessness of Quantum Gravity: I. The Evidence from the
Classical Theory'. Classical Quantum Gravity 11(12), 2853–73.
Barbour, J. (1994b) 'The Timelessness of Quantum Gravity: Ii. The Appearance of Dynamics
in Static Configurations'. Classical Quantum Gravity 11(12), 2875–97.
Baron, S., Evans, P., and Miller, K. (2010) 'From Timeless Physical Theories to
Timelessness'. Humana Mente 13, pp. 35–60.
Bourne, C. (2006). A Future for Presentism. Oxford: Oxford University Press.
Brogaard, B. (2007) 'Span Operators'. Analysis 67(1), 72–79.
Button, T. (2006) 'There's No Time Like the Present'. Analysis 66(290), 130–35.
Colyvan, M. (2001). The Indispensability of Mathematics. New York: Oxford University
Press.
Deutsch, D. (1997). The Fabric of Reality: The Science of Parallel Universes and Its
Implications. Michigan: Allen Lane.
Field, H. (1980). Science without Numbers. Princeton, N.J.: Princeton University Press.
Forrest, P. (2004) 'The Real but Dead Past: A Reply to Braddon-Mitchell'. Analysis 64(4),
358–62.
Healey, R. (2002). Can Physics Coherently Deny the Reality of Time? In C. Callender (Ed.),
Time, Reality and Experience, (pp. 293–316). Cambridge: Cambridge University
Press.
Ismael, J. (2002). Rememberances, Mementos, and Time-Capsules. In C. Callender (Ed.),
Time, Reality and Experience, (pp. 317–28). Cambridge: Cambridge University Press.
Lewis, D. (2004). Tensed Quantifiers. In D. Zimmerman (Ed.), Oxford Studies in
Metaphysics, (pp. 3–15). Oxford: Oxford University Press.
McCall, S. (1996). A Model of the Universe: Space-Time, Probability, and Decision. USA:
Oxford University Press.
McKinnon, N. (1999) 'The Hybrid Theory of Time'. Philosophical Papers 28(1), 37–53.
McTaggart, J. M. E. (1908) 'The Unreality of Time'. Mind, New Series 17(68), 457–74.
Mellor, H. (1998). Real Time 2. London and New York: Routledge.
Minkowski, H. (1908) 'Space and Time'. Reprinted in Smart 1964. Translated by W. Perrett
and G. B. Jeffrey. Originally an address delivered at the 80th Assembly of German
Natural Scientists and Physicians at Cologne.
Price, H. (1997). Time's Arrow & Archimedes' Point: New Directions for the Physics of Time.
Oxford: Oxford University Press.
Putnam, H. (1967) 'Time and Physical Geometry'. Journal of Philosophy 64(8), 240–47.
Schaffer, J. (2009). On What Grounds What. In D. Chalmers, D. Manley and R. Wasserman
(Ed.), Metametaphysics, (pp. Oxford: Oxford University Press.
Sider, T. (2003). Four-Dimensionalism: An Ontology of Persistence and Time. Oxford:
Oxford University Press.
Sider, T. (1999) 'Presentism and Ontological Commitment'. The Journal of Philosophy 96(7),
325–47.
Sider, T. (2006) 'Quantifiers and Temporal Ontology'. Mind 115(457), 75–97.
Skow, B. (2011) 'Why Does Time Pass?'. Noûs forthcoming.
Tooley, M. (1997). Time, Tense and Causation. Oxford: Oxford University Press.
24
Download